Improve speed of simultaneous burning

vbimport

#1

Hardware/Software: Two Phiilips 8601’s flashed to Benq 1620’s with v. B7P9 fiirmware, Dell 8400 w/3.0 ghz processor, 2 g ram., RAID 1 250 g harddrives, Windows XP Home Edition, Nero Ultra 6 fully updated.

Using one burner, it takes a little over 6 mins. to backup a movie from the harddrive to a dvd with Nero. Not bad based on what I’ve read here. When using both burners simutlaneously, however, completion time zooms to 16 or 17 mins.

Is this length of time normal for this type of operation? It makes sense that more time would be required to burn two discs concurrently – what is the consensus as to the best way to make simultaneous burns occur at a similar speed as when a single disc is burned? Do I have to reconfigure the cabel setup in the tower somehow or is there a software work around that does the same thing?

Drives are both on the same cable, the master is the end, the slave is in the middle. If this question is answered already (how could it not be?!), could some kind soul point me in the right direction? I’ve done extensive searches with no luck.

Thanks for your help and kindness. :bow:


#2

You will need to split the drives so they are on seperate cables and controller channels. You will also need to invest in an IDE controller card for your HDD’s (unless you have one already) to remove them from using the same channels as the burners (this will slow the burners down as well)

The BenQ’s only run at UDMA33 so burning at 16x for DVD needs 22MB/s & 2 drives pushes it upto 44MB/s, you have exceeded the UDMA33 spec & you’ll get underruns using 2 drives on the same cable.

Use Nero (or another program that can support dual simultanious burners) and select use multiple recorders.

That should allow you to hit 6 mins for burning 2 identical discs. BUT USE GOOD MEDIA as if one drive slows down nero will slow them both down. It only runs as fast as the slowest drive.


#3

The time should be the same whether using one burner or 2.

You also shouldn’t have to separate the burners, you should be able to do this with both of them on the same channel, but every IDE controller is different. In order to successfully burn to 2 drives simultaneously and avoid problems, they must be identical drives burning at the same exact speed, same firmware and identical media. This is true whether they are on the same channel or separate channels. The potential problem when they are on the same channel is that if one drive is even 0.1x faster than the other, it will have to pause to let the other catch up, and then the IDE controller will freak out.

It’s also good to note that 2 drives burning the same thing is NOT twice the data rate as one drive, it’s the same data rate. That’s why it’s a problem.

Suggest you use Nero, set Nero to show the actual write speed, and watch the speeds and buffers all through the burn.


#4

not in my experience and that’s with 2 PX716s each a Master on its own IDE channel with no slaves and source being 2x74gb raptors in RAID-0. adds anywhere from 30s-1:30mins to the burn time compared with a single drive burn in NERO. although it could be that one drive has PowerRec kicking in and the other has to wait (but i’ve used same media from same spindle). too bad the NERO reg trick to show real burn rate doesn’t work for dual burner recording (for obvious reasons).


#5

Yes, one of the drives is slowing down. You’re probly right that it’s PowerRec kicking in. This would happen whether they’re on the same channel or not.
There may be a small additional time added for Nero to spin up both drives, but the actual burn time should be the same.
I forgot about Nero not showing actual speed on multiple drives, but one should still be able to assess what’s going on by watching the buffers.

It seems that no 2 DVD burners burn at quite the same speeds, and therin lies the problem. With the newer walking OPC strategies, it’s unlikely that you’ll ever get 2 drives to run at the same speed cause they’re always pausing to calibrate. Limiting burn speed to 4x would be a good step.


#6

If we are talking seperate drives on a single data cable, each drive MUST have its own data, THEY CANNOT SHARE COMMON DATA!!! therefore twice the data must flow down the cable with a seperate packet going to each drive - therefore if you use 21MB/s per drive for 16x DVD writing on a channel able to transfer 33MB/s you WILL be trying to fit 42MB/s down it and run out of room when you try and supply 2 drives with the data.

 EACH DRIVE HAS TO HAVE ITS OWN DATA FEED - THEY CANNOT SHARE A SINGLE DATA PACKET BETWEEN THEM BOTH.

This is not a problem when there is enough bandwidth (e.g. if both units use UDMA66) but the BenQ is only UDMA33 therefore there is not enough bandwidth for 2 drives when writing above 8x for DVD. Remember that the channel must wait for each drive to acknowledge the packet before handing the channel back so if both drives use the same transfer rate the channel works at that rate. The channel does not run at UDMA66 because you have 2 UDMA33 drives attached. This is the main problem with the master / slave arrangment of IDE. If 2 drives transfer at max speed using the same transfer rate you get half the bandwidth per device.

 BenQ's 1620's MUST go on seperate channels for use above 8x DVD writing when using 2 drives simultaniously. 

If you want to test this, burn 1 disc at 8x and note the speed, if you now burn 2 at 8x you should se a similar speed as for just the 1.

 CD writing does not even get above 5MB/s so it has never had this problem.

#7

qwakrz,
Thanks for your response, all comments are most appreciated. If I understand you correctly, the easiest thing I can do to improve simultaneous buning performance is to lower the write speed to 8x, right? For the record, I’m using Ritek/Ridata 16x DVD+R (R04-01) media in both burners.

Pushing forward on the subject of changing the IDE cabel configuration inside the tower, I’ve looked again at the setup, and there are only two plugs for the floppy and the burners, the burners on one, the floppy on the other. The plugs on the circuit board they all attach to look about the same size, for what it’s worth. Talking me through the conversion from a master/slave arrangement to two masters may be beyond the scope of this format, at least for me.

Regardless, where would the IDE controller go, and how would my plug arrangement work. Do I have to disable my floppy?

Basically, having never attempted this manuver before, I would need step by step instructions in order to pull it off, you know? What do you thiink. . .


#8

drpino,
I’ve essentially asked qwakrz this question already, but I would like your two cents as well. How did you accomplish the setup you mention above? Thanks much.


#9

rdgrimes,
Interesting thought. . . thanks for the input. I forgot to mention, I’m using identical media (RITEK 04), and I am not making copies. The drives are identical as well. The goal is to siimultaneously burn different movies on different discs faster than 16 - 17 mins., keeping the quality at an acceptable level. When (and if) you do this operation, what burn times do you usually see?


#10

Your Benq drives use a WOPC strategy that means they will pause and re-calibrate periodically. And they will do this at different times. This fact alone might mean that you will be unable to get a smooth simultaneous burn whether they are on the same channel or different channels. (the other drive will have to pause at the same time) But the odds of success might be better on different channels. Switching to 4x (CLV) might eliminate this problem, but then you’re looking at 20 min times anyway.

This is an interesting topic, as nobody has really addressed the effects of the walking OPC on simultaneous burns. Each drive decides for itself how often it will pause and re-calibrate, so there might not be a good way to keep them in synch. If you were to put the drives on separate channels, and if your HD was able to supply the needed datarate, you would likely see faster times by burning from 2 processes at 8x. But it’s unlikely that your HD could keep up.

If you were trying to burn at 16x :eek: then the first step would be to try again at 8x, there will be less of the pauses and the time might well be faster than at 16x.


#11

@rdgrimes: thanks for your insight. i’ve tried dual burns at 16X and both drive buffers seem to be full, but they calibrate to each other after the first 30 seconds or so. like you said, i chalk it up to spin up time and PowerRec.

@qwakrz: thanks to you too for the info.

@Scuba Sam: not sure what motherboard you have, but every one i’ve ever seen has 2 IDE channels and not just 1. Usually 1 is black and the other blue, but that varies from mobo to mobo. you shouldn’t have to disable the floppy as that doesn’t run off an IDE channel but instead has it’s own dedicated socket on the mobo. im lucky enough to have SATA HDs in both machines so i can dedicate a burner to its own IDE channel. your configuration may not allow you this flexibility as your HD(s) may occupy one of the channels.

you mention you your goal is “to siimultaneously burn different movies on different discs faster than 16 - 17 mins”. unless you have a Hyper Threading Pentium 4 or dual processors, i wouldn’t recommend burning DIFFERENT movies at the same time. what we’ve been discussing is burning the same movie to 2 drives simultaneously. i just don’t think your CPU and/or HDs would be capable of this. if you think about it, burning 2 DIFFERENT movies CONSECUTIVELY at about 6:30 will net you 13:00 total which would achieve your goal of being under 16-17mins. you may be taking the hard route here.

i’ll keep track of this thread, as im interested in this topic as well.

thanks again to all you guys for your knowledge and perspectives.


#12

I have a Raptor RAID-0 array that will burn 2 different movies at 8x to which ever drives I choose. But 8x seems about the limit, 12x plus 8x might be possible, never tried that. CPU utilization is not much of a concern. I also burn movies while compressing another movie with 100% CPU use, but I don’t think it’s a very good idea.

Also have 2 Seagate Baracuda 160’s in RAID-1 that will not burn 2 separate movies at more than 4x. Although I’ve done it at 8x plus 8x, there are quite a few under-runs. Of course, burning 2 movies one from each array is not a problem at any speeds, CPU spikes to around 50% when doing that.

Personally, I wouldn’t burn any movie at 16x, the speed gains aren’t worth the potential loss of quality to me. 8x is plenty quick.

Anyway, I thought we were talking just about simultaneously burning the same thing to 2 drives.


#13

hrm, maybe i should try burning 2 diff things at the same time, thanks for that info rdgrimes.


#14

-Everybody,
Apologies for not being more clear. In my mind “simultaneous burning” meant “simultaneous buning of different discs.” I’m an English major and a bit of a writer as well so I’m totallly embarassed. Oh well, it was late. . . at least we’re all on the same page now.

I tried burning two different discs at 8x and it took 15:51, a slight improvement over what I had been seeing previously when burning at 16x. Regardless of the time, they view welll on my dvd player so I’m happy. I’ll post scans from those two discs later so we have some data to chew on. I also plan on doing a simultaneous burn test of the same disc (copies) just for the record.

I still think it’s odd how much longer burning two diffenent discs takes on my system when compared with burning one, but if that’s normal for my configuration, there’s not much else to say. Rearranging the IDE cables, however interesting it may sound, is probably overkill.

-drpino,
Great points all the way – thanks. Interestingly enough, I do have a Pentium 4, 3.0ghz w/HT.

-rdgrimes,
Quality trumps speed every time, you’re absolutely right. And, 8x is quite reasonble, especially if the quality is high.


#15

If your burn is not completing in the specified time, then you are getting many buffer under-runs. this is why I suggested watching the drive buffer (and read buffer) during the burn to confirm this. Allowing those buffer under-runs is asking for trouble, every once in a while you will get a failed burn. Your burn time at 8x should be around 8-9 min per disc max. Nero will report buffer under-runs in it’s log, but you should be watching to see when they occur.
I seriously doubt that your HD is capable of delivering enough data for 2 concurrent and separate 8x burns on the Benq drives.


#16

my thoughts exactly, as expressed earlier.


#17

8X…10 to 11 min
12X…7 min
16X…6 min


#18

can you elaborate furballi? are those your total burn times when burning 2 discs simultaneously at those speeds? i highly doubt you can even burn 2 discs at the same time at 16X or 12X.


#19

I tried running Nero analysis on two different discs burned together at 8x. The burn time was about 16 mins. Despite several attempts, CD/DVD Speed couldn’t read them, instead producing an error message: “Error - No Sense Info,” about half-way through the process. The discs view fine, and I’ve conducted quality tests on other discs so I know the program works and is installed correctly on my computer. As I produce them, I will continue to test different discs burned at the same time to hopefully get some postable scans.

-rdgrimes,
The read buffer went down to zero for a second or two during the aforementioned burn, confirming your previous post to be 100% accurate. When the upper big green bar in the Nero burn window goes to zero, this action constitutes a buffer underun, right?

Surprising that the discs finsihed burning and play great. . . even w/o the scans for verification, I think it’s clear that my system, as configured, is simply not up to the task of a concurrent, different disc burn.


#20

For fun, I just did 2 concurrant burns, one at 16x (CAV) and one at 8x (ZCLV). The 16x burn was finished before the other ever got to 8x, but both completed with no buffer underruns. :bigsmile: Gotta love those Raptor drives. :iagree:

Scuba Sam: It’s normal to see your drive buffer drop to zero a few times during the burn for a second or so. This is the pause initiated by the drive’s OPC scheme. Read buffer, however, should remain full at all times.